What Price Air?Mark Lewyn
Australian bureaucrats thought they had struck it rich last April when they auctioned off rights to use the airwaves for a satellite-TV service. An investment group called Ucom Proprietary Ltd. submitted a sealed bid for $152 million, far above government projections. The problem was that Ucom had no intention of spending that much. It deliberately defaulted on its winning bid, forcing the government to turn to the next lower bid--which was also Ucom's. The company proceeded to default on one bid after another before finally sticking with one that was just slightly higher than the bid of its closest rival, which by then had leaked out. Ucom walked off with the license for $84 million--setting off an uproar that nearly cost Australia's communications minister his job.
The U.S. Federal Communications Commission is determined not to make such a mistake later this year when it hands out licenses for new wireless phone systems known as Personal Communications Services (PCS). In the past, the agency awarded licenses free of charge, either through lotteries or hearings. This time around, it intends to hold its first-ever auction. To make sure the sale goes smoothly, the FCC has enlisted the help of an expert in game theory, John McMillan, an economist at the University of California at San Diego. His job is to apply the principles of game theory--a branch of applied mathematics that studies how people try to outsmart one another in contests ranging from checkers to politics--to help the FCC get the most for its licenses.
Not to be outdone, potential bidders have signed up their own game theorists. Pacific Bell has hired Stanford University Professors Robert B. Wilson and Paul R. Milgrom. Bell Atlantic Corp. has engaged Yale University Professor Barry J. Nalebuff and Stanford's Jeremy Bulow. MCI Communications Corp. has enlisted University of Maryland economist Peter Cramton. The consultants have been arguing in filings to the FCC over what the rules of the auction should be. Once the agency decides on the rules--which Congress has said it must do by Mar. 8--the theorists will change gears, helping their clients devise bidding strategies to exploit the system. If all goes as planned, which is by no means assured, the Office of Management & Budget hopes the auctions will raise more than $12 billion.
TRUE FUN. Game theory dates back to the 1940s, when mathematician John von Neumann and economist Oskar Morgenstern turned their formidable analytic abilities to playing their favorite game--poker. Their goal was to understand the best strategy to pursue--bluff, fold, stand pat, or raise--given a particular hand. Game theory was funded in the 1950s and 1960s by the Pentagon, which was interested in the best strategies for avoiding--and fighting--a nuclear war. But it has generally been shunned by business because the real world may be too complex and fickle to structure and analyze in any meaningful way. An auction, with rules known by all parties, is an exception, however. "This is one of the biggest applications of game theory in a long time," says University of Texas economist R. Preston McAfee, who is advising Airtouch Communications, the wireless business that is being spun off by Pacific Telesis Group.
The bidders' game theorists are having a rollicking good time devising strategies that will play to their clients' strengths--and minimize their weaknesses. Before its Feb. 28 announcement that it would take a 17% stake in Nextel Communications Inc., which has wireless franchises across the nation, MCI had lobbied the FCC to award a PCS license that covered the entire country. MCI figured that most other big telecom companies would be disqualified from bidding for such a broad license on the grounds that they already own substantial cellular properties. Furthermore, MCI proposed that the bid be awarded through a Vickrey auction, a sealed-bid procedure in which the high bidder wins but only pays the price of the second-highest bid (table, page 48). MCI calculated that with the pool of bidders small, the second-highest bid would probably be low. "It was so ridiculously self-serving, it wasn't funny," says Yale's Nalebuff. MCI later dropped the idea.
The FCC sees right through such proposals, which would limit the government's take. The agency does plan to assist small businesses, women, and minorities by setting aside some licenses on which only they will be allowed to bid--and by letting them pay in installments. But big businesses won't get any price break, though they complain that they'll have to spend so much for licenses that they won't have anything left to build their systems. Kent Y. Nakamura, an FCC attorney, argues that even if the government did devise an auction that left more money in the pockets of bidders, there's no assurance that they would invest it in wireless infrastructure instead of, say, executive bonuses.
To maximize the take, the agency will most likely go with a standard English auction in which the winner pays what it bids--and everyone can see all bids as they are being made, either electronically or in person. Game-theory research has shown that open auctions tend to stimulate bidding because the contenders don't worry as much about the "winner's curse" of overpaying: Sealed bids foster restraint for fear of needlessly making an offer that's way above the second-highest bid.
"GARBAGE BIDS"? The FCC also wants to get PCS to the public as rapidly as possible, however, which raises another issue: Should it take bids for all 2,500 or so airwave licenses at once? The agency intends to sell two licenses apiece in each of 51 regions called "major trading areas" and five licenses apiece in each of 492 smaller "basic trading areas." One camp of game theorists argues that all licenses should be put up for bid simultaneously. That would make it easier for buyers to assemble licenses into economically efficient blocs of adjoining areas. This method appeals to the FCC, because if bidders get what they need at the auction they won't waste time trading licenses afterward, and wireless services would reach market more quickly.
Trouble is, requiring buyers to submit all their bids at once could be a nightmare of complexity. And the more complicated the auction is, the greater the threat of legal challenges from losing bidders--a major concern of the FCC. Beyond that, the rules would probably have to allow contenders to withdraw and shift around offers--so a bidder who lost New York wouldn't be stuck with paying a high price for a contiguous New Jersey property that it no longer wanted. But if offers can be withdrawn, "people can submit all sorts of garbage bids," says Northwestern University economist Daniel R. Vincent, who is advising Washington-based American Personal Communications Inc., a company backed by Washington Post Co. Without serious bids as a guide to value, buyers tend to be cautious--and come in low.
BIG UNKNOWNS. Of course, the alternative isn't entirely palatable, either. In a sequential auction, using either sealed bids or open ones, licenses are put up one at a time, like paintings. Because buyers can't go after geographic blocs all at once, they may not bid as much for properties. "The problem with sequential is that it doesn't allow bidders to fully express what they might be willing to pay," says the FCC's Nakamura. Bell Atlantic wants a sequential auction because it is less complex and less subject to legal challenge. But Airtouch Communications, the Pacific Telesis spin-off, doesn't like this idea because such an auction might drag on and on, wasting everyone's time. If the auction is sequential, Nynex Corp. is hoping that properties won't be auctioned off from biggest to smallest--because it doesn't want New York to go first. Game theory suggests that Nynex might be able to save money by calibrating a bid for New York from the winning bids for other areas.
Academics have devoted entire careers to exploring just such questions. As the Mar. 8 deadline approached, the FCC still had to make major decisions, including whether to choose simultaneous bidding, sequential bidding, or some hybrid. "There is no obvious best way," says Yale's Nalebuff. One thing is clear, though: No matter what rules prevail, the government is bound to do better than it has in the past. Since the dawn of broadcasting, the FCC has been making gifts of licenses. It doesn't take a game theorist to point out that any winnings are better than none.